Conference Draft Not for Quotation Long-Term Objectives for Government Debt Alan J. Auerbach University of California, Berkeley January 2008 This paper was prepared for presentation at a conference on Fiscal Policy and Labour Market Reforms organized by the Swedish Fiscal Policy Council, Stockholm, January 29, 2008. I am grateful to Bill Gale and Kent Smetters for comments on an earlier draft.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Conference Draft Not for Quotation
Long-Term Objectives for Government Debt
Alan J. Auerbach University of California, Berkeley
January 2008
This paper was prepared for presentation at a conference on Fiscal Policy and Labour Market Reforms organized by the Swedish Fiscal Policy Council, Stockholm, January 29, 2008. I am grateful to Bill Gale and Kent Smetters for comments on an earlier draft.
1. Introduction
Governments use current levels of national debt, typically scaled by contemporaneous
GDP, along with the annual change in national debt – the budget deficit – as measures of
fiscal position. But what should government policy aim to achieve with respect to these
measures? And, are these the right summary measures at which to be looking? This paper
considers what the government should use as its fiscal targets to achieve policies that are
consistent with long-term fiscal objectives. When focusing on long-term objectives, though,
one must keep in mind the interaction with shorter term objectives. A key issue that arises,
for example under the Stability and Growth pact, is how long-term objectives should be
modified to accommodate economic fluctuations and to maintain some flexibility with respect
to fiscal stabilization policy.
The analysis proceeds in three parts. The next section of the paper discusses the
underlying long-term objectives that lead to a focus on debt and deficits, and points out that
these objectives may call for different fiscal targets, the relative importance of the different
targets depending on both social judgments and economic behavior. Section 3 offers an
analysis of the difficulties encountered when one attempts to use simple rules based on debt
and deficits to produce policies that are consistent with these long-term objectives. In light of
these difficulties, Section 4 considers a variety of alternative measures that might serve better
the goal of aligning fiscal targets with underlying long-term objectives. Section 5 offers some
brief concluding comments.
2. Long-Term Fiscal Objectives
There are at least three important long-term objectives that appear to be associated
with concerns about debt and deficits: intergenerational equity, economic performance, and
- 2 -
fiscal sustainability. Unacceptably high levels of debt are seen as compromising all three
objectives.
A. Intergenerational Equity
A simple view of national debt is that it is an obligation that is passed from current
generations to future generations. Thus, the higher the national debt, the more future
generations will have to pay. Based on a view of social welfare that takes account of the
relative well-being of current and future generations, one presumably determines the level of
national debt at which the burdens being passed to future generations produce the most
equitable distribution of resources among generations. Unlike the other underlying objectives
to be discussed, intergenerational equity cannot be evaluated by economic analysis alone,
because it requires judgments about how to weigh the welfare of different individuals.
B. Economic Performance
There are several channels through which national debt is viewed as hampering
economic performance. First, to the extent that it produces a perception of increased private
wealth among those who hold it, national debt may reduce the accumulation of real assets
through the “crowding out” process. This reduction in saving leads either to reduced domestic
capital accumulation or reduced foreign capital accumulation, i.e., an increase in net inbound
capital flows and a corresponding worsening of the current account.
Second, allowing debt to accumulate may necessitate a future increase in revenues as a
share of GDP and an associated increase in marginal tax rates on economic activity. Given
that the economic cost of tax distortions rises roughly with the square of tax wedges,
fluctuations or trends in tax rates are undesirable and a smoothing of marginal tax rates over
time will generally minimize the deadweight cost of taxation (see, e.g., Barro 1979).
- 3 -
Third, increased debt may ultimately force the government into a faster rate of money
creation to meet its fiscal obligations, with the consequence of higher inflation and an increase
in the associated distortions and disruptions (Sargent and Wallace 1981).
C. Fiscal Sustainability
Future government policies are tied to the current fiscal situation through the
government’s intertemporal budget constraint, which says that the stock of current net
liabilities must equal the present value of future primary budget surpluses. This budget
constraint must hold, in the end, if we rule out permanent “Ponzi games” in which the
government continually issues debt at a rate sufficient to avoid having to raise any resources
for debt service. But how the intertemporal budget constraint is ultimately satisfied is not
determined by the constraint’s mere existence.
If the level of government debt and the projected path of primary surpluses indicate
that current fiscal policy is not sustainable, then a variety of things might occur to alter this
path. There may be voluntary changes in monetary and fiscal policy, such as tax increases or
money creation, which bring with them the undesired economic consequences just discussed.
But changes in fiscal policy may also be precipitated by a fiscal crisis that brings with it
considerable economic disruptions. There might even be increases in the price level prior to
any monetary response, according to the “fiscal theory of the price level,” which posits that
the price level, rather than fiscal policy changes, serves to balance the government’s
intertemporal budget constraint (see, e.g., Woodford 1995).
Thus, the issue of sustainability relates to that of economic performance, but with a
focus on the additional economic costs, beyond those associated with fiscal policy itself, that
may be experienced as a result of having to adjust fiscal policy precipitously away from an
errant trajectory. That is, there may be significant costs of adjusting policies that are far from
being sustainable, in addition to whatever costs are associated with the new, sustainable path
- 4 -
or the transition from one path to another. It is useful to treat these additional costs separately,
as they depend on different factors than those already considered.
D. Weighing the Different Objectives
When debt and deficit targets are being determined, which of these different objectives
are most important depends on characteristics of the economy. For example, under the polar
case of pure Ricardian equivalence (Barro 1974), with each generation taking the well-being
of future generations fully into account in planning its saving and bequests, the level of debt is
of no consequence to intergenerational equity. With private saving and bequests being
adjusted to offset changes in debt, the level of debt conveys no information about the
intergenerational distribution of resources, nor does the deficit provide any information about
changes in this distribution.
Under the same assumption about generational connections, government debt loses
some of its macroeconomic effects as well. In particular, the wealth effects that increase
current consumption and induce crowding out of domestic or foreign asset accumulation
vanish, because the increase in government debt carries with it an increased liability of some
future generations which current generations treat as their own. In a Ricardian world, then,
the main purpose of debt and deficit targets is to determine a path along which economic
distortions can be kept to a minimum and which, if adhered to, can ensure a fiscal policy path
that is sustainable.
The relevance of the Ricardian equivalence proposition has been discounted in the
literature examining the consumption behavior of related family members (see, e.g., Altonji,
Hayashi, and Kotlikoff 1992). Even if the proposition does hold to some extent, the
implications for debt and deficit targets is not entirely clear, as the debt trajectory consistent
with intergenerational equity (ignoring offsetting private transfers) may be similar to the debt
trajectory that would provide a smooth path for tax rates. That is, we might want stable ratios
- 5 -
of tax revenue to GDP to spread burdens fairly as well as to smooth marginal tax rates. Still,
the objectives do not lead to identical target paths for debt.
For example, the objective of generational equity, when per capita incomes are rising
over time, might call for rising revenue-GDP ratios, just as progressive taxation at any given
time is viewed as an equitable response to within-cohort income differences. But a steadily
rising revenue-GDP ratio would (without a change in the tax structure itself) lead to rising
marginal tax rates and greater tax distortions than necessary if generational equity is
unaffected by the policy and hence not relevant to the choice of an optimal debt path.
Likewise, the desire to maintain debt on a stable path has similar but not identical
implications for debt targets as the objectives of intergenerational equity and economic
efficiency. A rapidly exploding debt-GDP ratio is inconsistent with all three objectives, but
one can imagine a stable fiscal policy with steadily rising revenues as a share of GDP, again a
path not consistent with the minimization of the deadweight loss from distortionary taxation.
In summary, a desired path for debt depends on the objectives that path is supposed to
achieve. It is difficult to imagine how one can arrive at debt targets without specifying these
objectives and determining their relative importance.
E. Targets versus Restrictions
Aside from specifying the objectives that underlie targets for debt, one must also
indicate the purpose of specifying this path. At one extreme, the debt targets simply convey
information in a complex setting to a benevolent government that seeks to achieve the
objectives that underlie the targets. In this case, once the information if provided, no further
mechanism is necessary – the benevolent government will implement the desired polices once
their identity has been revealed, and private agents will possess an accurate forecast of
government actions on which to base their own economic decisions.
- 6 -
A more likely situation, though, is that the government has conflicting objectives that
may lead it to seek to deviate from the socially optimal path. There are different reasons why
governments might wish to behave in a manner that is fiscally irresponsible when compared
to the target. First, if voters are myopic as to the future consequences of current policies, it
may be possible to manipulate election outcomes.1 Second, a government anticipating a
potential loss of power may seek to deny resources to future governments by making strategic
use of its current ability to commit future resources (Persson and Svensson 1989, Alesina and
Tabellini 1990).2 In the first case, the provision of information may help, to the extent that
informed voters may be less likely to be swayed by governments providing unsustainable
transfers, tax cuts, etc. But in the second case, excessive deficits have little or nothing to do
with imperfect information on the part of voters. Hence, depending on the source of a pro-
deficit policy bias, it may be that provision of information must be accompanied by some sort
of restrictions on fiscal decisions in order for the information to have the desired impact on
fiscal policy.
The distinction between providing information and imposing restrictions has
implications for the nature of the fiscal policy targets. Most importantly, it may be desirable
to provide less flexibility to governments that might wish to avoid meeting the targets. For
example, suppose that it would be optimal for the government to follow a policy that, on
average, achieves a deficit equal to a certain specified percentage of GDP, with the deficit-
GDP ratio varying with respect to certain economic conditions that may not be fully
measurable immediately, such as temporary weakness in the earnings of high-income
individuals that leads to a sharp drop in tax revenue. A government seeking to maximize the
deficit might take advantage of uncertainty regarding economic conditions by offering a
1 The literature on the determinants of voting (e.g., Fair 1978) suggests that voters base their decisions on very recent economic performance. Politicians may also deviate in attempting to provide signals to voters about their views or abilities (e.g., Rogoff and Sibert 1988).
2 Auerbach (2006a) discusses the design of budget restrictions in such a model.
- 7 -
biased prediction with regard to the prevailing state of nature. Thus, a simple rule requiring a
particular deficit-GDP ratio, while removing desirable flexibility from the government, might
also eliminate a pro-deficit bias.3
As another example, suppose that it is deemed desirable to reduce the debt-GDP ratio
by a certain amount over a ten-year period, with the exact path to depend on the strength of
the economy. A government claiming it seeks to avoid fiscal contraction might simply
announce a plan to accomplish the entire adjustment near the end of the ten-year period, so a
specified path of adjustment might be preferred, even though this would reduce the
government’s capacity to adjust to variations in economic circumstances over the period. In
short, if the government has superior information regarding the economic state of nature, then
we would like to allow it to use this information if its incentives align with social objectives,
but not necessarily if objectives diverge. The same conclusion applies even if the government
simply can claim to have superior information, i.e., it can argue in a manner that is difficult to
refute that its policies are consistent with the stated target.
Of course, if the problem with flexibility is due to the government’s being able to take
advantage of actual, or claimed, superior information, an alternative to reducing its flexibility
is to establish an independent entity that is privy to the same information but has no incentive
to have fiscal policies deviate from the target. This, presumably, is a central purpose of an
independent fiscal policy council, particularly if the council has no separate decision-making
role. In order to make this mechanism of independent fiscal evaluation work effectively,
though, some sort of penalty structure may be required as well, assuming that simply
announcing that the government has missed its target does not impose enough of a penalty on
its own.
3 An alternative here might be to allow flexibility but to impose an average target for the deficit-GDP ratio that is smaller than optimal to correct for the anticipated bias in policy.
- 8 -
F. Long-Term Fiscal Objectives: Summary
There are at least three distinct objectives that underlie the determination of long-term
fiscal policy targets: intergenerational equity, economic efficiency, and fiscal sustainability.
Though these objectives tend to push policy in the same direction, they are distinct and can
give rise to conflicts regarding optimal policies. Given the conflicts, it is relevant how
important each of these objectives is, with relative importance depending not just on social
attitudes (such as the strength of aversion to intergenerational inequality) but also on
economic behavior (e.g., the strength of the bequest motive or the responsiveness of economic
activity to high marginal tax rates).
If governments have incentives not to adhere to fiscal policy targets, then restrictions
on fiscal policy actions may be desirable, even though such restrictions reduce the scope for
varying policy in response to changes in economic conditions. An independent entity such as
fiscal policy council can serve as an alternative mechanism for ensuring compliance, although
a certification of non-compliance, alone, may not impose a sufficient penalty.
3. The Inadequacy of Simple Deficit Targets and Budget Rules4
The current level of the national debt measures the existing obligations that must be
met through future taxes in excess of future spending. The current budget deficit provides a
measure of where national debt is going. But neither the debt nor the deficit is an adequate or
unambiguous measure of the current state of fiscal policy, in terms of the objectives discussed
above, and hence neither is an ideal tool to use in establishing fiscal policy targets.
A. Government Assets
Governments accumulate assets as well as liabilities, and it has been argued for a long
time (see, e.g., Eisner and Pieper 1984) that a measure of a government’s debt or deficit that
4 Some of this analysis relies on the discussion in Auerbach and Kotlikoff (1987), chapter 4 and Auerbach, Gokhale and Kotlikoff (1992).
- 9 -
does not net out assets and asset accumulations provides an incomplete and biased measure of
the government’s fiscal policy position. Indeed, there are fiscal policy rules, for example for
most state governments in the United States (see, e.g., Poterba 1994, Bohn and Inman 1996),
that reflect this position, restricting or entirely prohibiting deficits for current spending but
allowing deficits for capital spending. The logic of subtracting assets from debt and focusing
on net assets is straightforward: just as debt is a burden on future generations, assets are a
benefit.
But there are at least two considerations that attenuate this symmetry. First, even if the
assets are owned directly by the government, they may not provide a revenue stream to the
government to offset the cost of servicing liabilities. For example, borrowing funds to build a
public park may have offsetting effects on future generations, but unless there is an admission
charge assessed by the park, tax revenues and, presumably, tax rates will still need to rise to
service the debt. Thus, assets might be properly netted against debt in constructing a measure
used to assess intergenerational equity, but this might be the wrong approach for considering
whether efficient tax smoothing is being achieved. Second, when does an expenditure item
qualify to be treated as an asset acquisition rather than a current expenditure? Much of what
the government spends money on, for health, education, etc., is intended to benefit individuals
beyond the current year. To the extent that such expenditures convey no future tax revenue,
one may wish to exclude these expenditures when computing assets, at least when thinking
about tax smoothing.
Note, though, that even if an expenditure generates no government revenue directly
(through admission charges, road tolls, sales by government enterprises, etc.), it may do so
indirectly. For example, an expenditure on education (or health) may increase the earnings
capacity of workers who, as a consequence, pay higher taxes in the future. The expenditure
- 10 -
may therefore represent an asset to the government even if the scope is limited to assets that
generate revenue.
This example provides an illustration of what has been referred to as “dynamic
scoring” in the debate about measuring the fiscal effects of taxes and spending (see Auerbach
2005, Altshuler et al. 2005), although the more common application has been in considering
the net impact on tax revenue of provisions that raise or lower taxes. For example, reductions
in marginal tax rates may spur economic activity and allow the government to recoup some of
the revenue lost as a consequence of the initial tax cut. If a component of government
spending is treated as an asset because it generates revenue in this indirect manner, then the
indirect revenue effects of tax provisions need to be counted as well in order to make the
calculation consistent.
In summary, although accounting for government assets makes sense, doing so is more
difficult than accounting for explicit government liabilities, because there is no precise
method of distinguishing assets from current expenditures and because assets, however
defined, vary in the extent to which they provide revenues that can be used to offset the
interest payments on the national debt. Counting the tax revenues that result indirectly from
expenditures may make sense, but only to the extent that the same treatment is given to tax
revenues that result indirectly as a consequence of changes in tax policy.
B. Implicit Liabilities
The government liabilities that enter into the calculation of government debt are
explicit liabilities of the government. We think of these debts as requiring future revenues
because we assume that the government will not default. Using this criterion, though, there
are many other government liabilities as well. Essentially all developed economies have
unfunded public pension schemes, meaning that future revenues will be needed to cover the
- 11 -
benefits that have already been promised to existing workers and beneficiaries. Likewise,
given public provision, the need to pay for health care represents a large government liability.
Some might argue against treating such future expenditures as current liabilities, on
the ground that they do not represent legal claims in the same way that explicit government
debt does. But it is an essentially empirical question whether the government is likely to meet
its future liabilities, and if so to what extent. Indeed, the explicit liabilities of countries at risk
of default often trade at a considerable discount while some public pension commitments have
proved difficult to adjust, so the idea that we should count explicit liabilities fully and ignore
other commitments entirely seems at odds with reality.
It is important to clarify that it is not the expenditures themselves, but the way in
which they are financed, that leads to the existence of implicit liabilities. Consider, for
example, a health care system that promises health care benefits to the elderly. If the elderly
are assessed user fees for these benefits, there is no reason to treat these future expenditures as
a current liability – there is no obligation to raise future taxes to cover them. If, on the other
hand, the expenditures are financed, say, through payroll taxes, then key elements of national
debt are present: the need for distortionary tax finance, coupled with a shift in the burden to
future generations. In this case, the only remaining question is the extent to which the
government has a commitment to make the future expenditures.
This logic largely applies even to systems that are self-financing, such as true pay-as-
you-go public pension schemes under which taxes or benefits are by law adjusted annually to
maintain cash-flow balance.5 Such systems still incorporate a liability to older generations
that must be financed by distortionary taxes on younger generations. The only aspect of
5 While many pension schemes are characterized as being of the “pay-as-you-go” type because taxes pay for current benefits, the reference here is to a system in which taxes and benefits are also equal. Such a system is currently in place in Germany, whereas Sweden’s system of Notional Defined Contribution accounts incorporates a balancing mechanism to implement adjustments over time should revenues and expenditures begin to diverge. See Auerbach and Lee (2006) for further discussion.
- 12 -
national debt apparently missing is the prospect of being on an unsustainable fiscal path, given
that taxes and/or benefits are continually adjusted to ensure sustainability.
But similar provisions can be introduced even to deal with explicit national debt,
requiring automatic cuts in spending or increases in taxes should the debt or deficit exceed
some acceptable level.6 Having such provisions in place does not make the debt itself
disappear. Also, having in place an automatic provision to adjust revenues or benefits does
not guarantee that such adjustments will be feasible. A public pension system with an
increasing dependency ratio could, in principle, get to a point where the maximum revenues
achievable, given labor supply responses to increasing payroll taxes, fell short of the benefits
to be financed, or where such high taxes were politically impossible to sustain.
In summary, the government has many implicit liabilities that have some or all of the
important characteristics of explicit debt liabilities: commitment, intergenerational transfer,
the need for distortionary tax finance, and the possibility of being unsustainable. A key
difference between explicit and implicit debt, though, is the heavy dependence of the latter on
demographics.
Whereas explicit liabilities have a specified value, implicit liabilities are created by
government programs that typically specify levels of benefits and taxes rather than of
liabilities. As dependency ratios change over time due to changes in labor force participation,
fertility, mortality and morbidity, implicit liabilities change as well. Measuring the value of
implicit liabilities, therefore, requires making predictions of the future values of these
variables. While these predictions are subject to considerable uncertainty, the range of
possible outcomes can still be combined to produce a single liability estimate. A related
question is over how many generations the implicit liabilities should be computed, in
6 Such a system was in place in the United States in the late 1980s under the Gramm-Rudman-Hollings legislation, which called for a process of “sequestration” under with automatic budget cuts would occur if a year’s deficit target were missed.
- 13 -
particular whether to include liabilities to all generations (so-called “open-group” liabilities)
or liabilities just to existing generations, or to existing generations above a certain age (so-
called “closed-group” liabilities). If a measure of the liability being passed on to future
generations is the objective, then the closed-group liability may be the most natural to use.
But the open-group measure may be of use as well, particularly concerning the question of
sustainability.
C. Tax Structure
In constructing a measure of the budget deficit, we subtract revenues from
expenditures. But not all revenues have the same impact on economic efficiency or the same
generational burden. Just as one may identify implicit liabilities in spending programs that
have the same attributes as explicit national debt, there are differences among tax systems that
amount to the establishment of implicit assets and liabilities. Ignoring these differences
makes no more sense than ignoring explicit government liabilities.
Consider, for example, the difference between payroll taxes and consumption taxes,
such as the value added tax (VAT).7 From the perspective of any given taxpayer, the
consumption tax differs from the payroll tax in two important respects. First, it is collected
later in life. Second, it has a broader base, from a lifetime perspective, because it falls not
only on consumption financed by employment earnings, but also on consumption financed by
previous accumulations of wealth. Thus, at any given time, if one looks forward, the taxes
that will be collected from any individual will differ under the two tax bases. Consumption
taxes will be higher than payroll taxes, for two reasons. First, taxes associated with
consumption financed by wages will be collected when the consumption occurs, rather than
when the wages were earned. Second, taxes will also be collected when consumption
financed by prior asset accumulation occurs. Thus, if we use the payroll tax as a benchmark,
7 This discussion follows that in Auerbach (2006b).
- 14 -
the consumption tax incorporates an implicit government asset in the form of a right to receive
future tax payments.
To make this analysis concrete, consider the important case of the taxation of private
pensions. Typically, when funds are set aside for private pensions, the contributions to
pension funds are deducted from wage compensation, so that they are not subject to current
taxes on wage income. After these funds accumulate and are paid out to retired workers, they
may be taxed. This treatment of deferring taxes on pensions amounts to converting labor
income taxes into consumption taxes, in that withdrawals from pension funds by retirees are
for the purpose of consumption.
Even if this effective conversion of labor income taxes into consumption taxes has no
impact on the present value of taxes collected, it does defer their collection. Thus, we might
say that the government has a claim on a portion of pension fund accruals, as it expects to
receive a portion of such accruals once they are distributed. This makes sense because,
relative to a system under which the initial pension contributions rather than the distributions
were taxed, there is no difference in the generational burden of taxation, and no difference in
the economic distortions associated with taxation. Indeed, if we placed a market value on the
pension fund from the beneficiary’s perspective, the fund would carry a discount equal to the
present value of the deferred taxes, the government’s deferred-tax asset.
Just as with implicit liabilities, the size of deferred tax assets depends on demographic
factors. For example, a country with an aging population and a maturing public pension
scheme might have a rapidly accumulating claim to taxes on pension fund withdrawals.
Indeed, some analyses of budget sustainability have pointed to this dependence on
demography in arguing that such deferred tax assets may help to offset the rising implicit
liabilities of old-age transfer programs.8
8 See Auerbach, Gale and Orszag (2004), who argued that, at least for the United States, the value of deferred-tax assets falls far short of the value of implicit liabilities.
- 15 -
While the example of deferred taxes on pension saving is particularly straightforward,
the analysis generalizes to the tax system as a whole. Once we have established a baseline tax
structure, we can identify deviations from that tax structure that give rise to deferred tax assets
and calculate what these assets are. Using a payroll tax as a benchmark, for example, we
would include the deferred taxes on saving, as just described, and add the taxes on
consumption from previously accumulated wealth. Similarly, business level taxes that hit the
returns to existing assets more heavily than returns to new assets effectively impose deferred
tax liabilities on existing assets that will be capitalized into existing asset values and can be
quite large under existing tax systems.9
As with implicit liabilities, deferred tax assets share key properties with explicit
government liabilities, although in the case of deferred tax assets the effects are in the
opposite direction. First, deferred tax assets represent a generational transfer: the taxes due
from future pension recipients reduce that need for contemporaneous collections from other
future taxpayers. Second, in reducing the need for tax collections, deferred tax assets reduce
the need for further tax distortions. Third, the existence of deferred tax assets makes fiscal
policy more sustainable. And, as in the case of implicit liabilities, the existence of deferred
tax assets depends on the strength of government commitments, in this case to maintain the
expected tax structure. By analogy to the case of implicit liabilities, though, we can take
account of the range of possible future tax rates in computing the current value of deferred tax
assets.
The calculation of deferred tax assets must start, though, from the specification of a
baseline tax structure. Relative to a payroll tax, a consumption tax incorporates deferred tax
assets, as does the partial consumption tax treatment introduced by the deferral of taxes on
private pension accruals. However, if one chose a consumption tax as the benchmark tax
9 See, for example, the calculations in Auerbach (1983, 1996) for the U.S. corporate tax.
- 16 -
system, then a consumption tax would incorporate no additional deferred tax assets, and a
payroll tax would instead involve deferred tax liabilities – the prepayment to the government
of taxes on future consumption, with the government liabilities coming due when
consumption occurs in the future without any additional taxes being paid.
The same need for a benchmark can be seen in the case of implicit liabilities. If we
take the existence of a public pension system of a given size as part of the “standard” tax-
transfer system, then only increases or decreases in commitments will give rise to implicit
liabilities (or implicit assets, in the case of a reduction in commitments). In both cases,
deferred tax assets and implicit liabilities, there may be a natural or useful benchmark from
which to start (for example, one in which fiscal policy involves no intergenerational transfers,
with taxes and spending aligned), but the benchmark must be specified before the asset and
liability adjustments can be made.
D. Changes in the Desired Path of Deficits
Leaving aside the difficulty of measuring assets and liabilities, there are a number of
reasons why the desired path of debt and deficits may change over time. One, already
mentioned, is the need for short-term stabilization policy. Due to the automatic stabilizers
built into the tax and expenditure system, the deficit will react to the state of the economy,
even with no explicit change in policy. Offsetting these changes would amount to pro-
cyclical fiscal policy that could exacerbate economic cycles.
But even a passive response may be viewed as inadequate, to the extent that the
government believes it can adjust taxes and spending quickly enough to lessen economic
fluctuations through variations in fiscal stimulus. If active fiscal policies are seen as
desirable, then a deficit target will need to reflect this, being sensitive enough to the cycle not
simply to permit automatic stabilizers to work, but also to make room for new policy actions.
As discussed above, though, providing this flexibility may be difficult when there is imperfect
- 17 -
information about the state of the economy, particularly if a government’s incentives push in
the direction of increased deficits.
It is a standard and uncontroversial argument, of course, that debt should be allowed to
accumulate during times of war or when other conditions exist that cause the need for
resources to be temporarily high. While wars are easily observable, though, it may be more
difficult to judge the extent to which other circumstances present an argument for debt
accumulation.10
Formally, the rationale for allowing increased deficits in times of war or other
emergencies is that the valuation of spending during these periods is unusually high; thus,
higher spending may be justified. From a tax-smoothing smoothing perspective, it is better to
let debt accumulate rather than to have tax revenues keep pace with the sudden surge in
spending. From a generational equity argument, too, it is easy to justify debt, given that the
added spending typically does not diminish the value of providing resources to cohorts present
when the added spending occurs. That is, individuals experiencing wars or natural disasters
do not have a higher ability to pay than other generations as a consequence of the surge in
public spending that accompanies such events. Indeed, a government seeking to smooth the
marginal valuation of resources among generations may wish to direct even more resources in
the direction of such generations.
While wars and emergencies are obvious occasions to modify debt targets to allow
increases in spending, any variations over time in the value of marginal spending should have
the same effect. These variations can be due to changes in technology, for example with the
increase in automobile usage increasing the value of additional road construction. But an
important factor, particularly as one looks ahead, is demographic change. An increase in the
10 During the late 1990s in the United States, the Budget Enforcement Act then in effect placed limits on discretionary spending by the federal government. The law provided an exception for emergency spending. As budget surpluses accumulated, so did political pressure to increase spending, with the result that there was a surge in what was classified as emergency spending, including many expenditures that were of a recurring nature and easily anticipated. See Congressional Budget Office (1998).
- 18 -
population of school-age children increases the social value of additional spending on
education, while an increase in the share of the population that is elderly may increase the
value of additional spending on medical care. A simple rule, consistent with the fiscal
response to emergencies, might be to adjust deficit targets to keep age-related spending
constant in per capita terms as the population’s age structure changed, thereby keeping the
valuation of additional spending constant. But this rule would not be optimal if such changes
were of a permanent nature, for there would be no mechanism for offsetting the resulting
change in spending over time. For example, an increase in the deficit in response to a
permanent increase in the elderly share of the population would simply result in an
unsustainable path of accumulating debt.
E. Deficits versus Spending
A question that often arises in establishing fiscal targets is whether the targets and/or
restrictions should apply to deficits or to spending. There seem to be at least two arguments
in favor of applying restrictions to spending rather than to deficits, i.e., the difference between
spending and revenue.
The first argument for focusing on spending relates to the notion of an asymmetry
between spending and revenue. Whereas an increase in spending represents an exhaustive use
of resources by the government, a reduction in taxes that has the same impact on the deficit
does not. Thus, so the argument goes, a spending target is needed to ensure that resources
used for marginal public spending provide a social value no lower than in private use.
A second, related argument is that spending is a measure of the size of government,
whereas the deficit is not. A deficit of a given size can correspond to high or low levels of
spending and tax revenues, so a desire to maintain a small government, perhaps to maintain
the strength of individual liberties, cannot be accomplished through deficit control alone.
- 19 -
While there may be some merit to each of these perspectives, there is also a very
serious problem with each, namely, that whether a particular budget item represents an
increase in spending or a reduction in taxes may be largely a matter of form rather than of
substance. This ambiguity has been recognized at least since the concept of “tax
expenditures” was introduced (Surrey, 1974), illustrating how programs to accomplish
particular objectives could be organized through the tax system via reductions in the tax base
rather than as direct spending programs.
While this ambiguity is perhaps most evident in the case of transfer payments, which
are sometimes netted against taxes and in other cases added to spending, tax policies that
encourage home ownership, charitable contributions, and other private activities could also be
accomplished through direct government subsidies, and direct spending by government on
items such as education could be replaced by private spending encouraged through the
provision of tax incentives.
If government spending is not a well-defined concept, then a mechanism targeting the
control of government spending (or, for that matter, the control of government revenues) is
poorly conceived. A government that uses the tax system to accomplish its objectives can be
just as intrusive as one using direct spending, and the use of the funds may be identical.
However, it may be that some elements of government spending are difficult to accomplish
through the tax system, and if so it may be appropriate to include a combination of spending
and deficit measures in the portfolio of fiscal targets.11
F. Deficits and Generational Distribution
A final difficulty of using annual fiscal measures, whether deficits or spending, to
assess the change in fiscal climate is that such measures do not provide enough information in
11 This was the approach taken under the previously cited U.S. Budget Enforcement Act, which placed annual limits on discretionary spending but also limited the extent to which legislation could increase the overall budget deficit.
- 20 -
relation to the fundamental objectives of fiscal policy. Adjustments for government assets,
implicit liabilities, or deferred taxes all represent attempts to achieve a measure that comes
closer to what we want, but these adjustments do not perfectly address the problem. An
example will illustrate why problems remain.
Suppose that the government is considering three changes in its existing unfunded
public pension system, which we will assume to be run with annual cash-flow balance and
financed by payroll taxes. Table 1 summarizes these three policy changes and their various
effects. The first policy would increase the size of the pension system by increasing payroll
taxes beginning in five years and increasing the subsequent benefits of those paying higher
taxes by an amount equal in present value. The second policy would reduce pension benefits
for all recipients beginning in five years, by the same aggregate annual amount as the tax
increase under the first policy, with no change in payroll taxes. The third policy would begin
reducing benefits starting in twenty years, with the same present value reduction in benefits to
current generations as the second plan, i.e., reducing benefits by less initially but eventually
by a greater amount.
None of these plans would have any impact on the current national debt or deficit, as
shown in the column 1 of the table. The first two plans would also have the same impact on
the annual budget deficit five years out and for several years in the future and hence the same
trajectory of explicit national debt (column 2), providing the picture that both eventually
reduce the national debt. But these two plans would have fundamentally different economic
consequences.
The first policy has no impact on current generations, old or young (columns 4 and 5)
for it would maintain the benefits of older generations and would increase benefits in line with
taxes for younger generations. Hence, despite its reducing the budget deficit over the short
term, this policy would have no impact on the system’s implicit liabilities to existing
- 21 -
generations (i.e., the closed-group liabilities, column 3) and hence no change in the debt being
passed on to future generations (column 6).
The second policy, by reducing benefits of current generations, both old and young,
would reduce the burden being passed on to future generations. Including the pension
system’s implicit liabilities to current generations in a measure of national debt would account
for the difference between the first and second policy, recognizing that the first policy results
in higher implicit liabilities and higher future marginal tax rates than the second.
The third policy has no impact on the budget deficit, even over the first twenty years
but, like the second policy, it reduces the burden being passed to future generations.
Accounting for implicit liabilities to existing generations reflects this reduction accurately.
But this fiscal policy has different effects on existing generations than does the second policy,
even though they produce the same reduction in implicit liabilities. By delaying
implementation, it allows current retirees to escape benefit cuts, which then must fall harder
on younger generations, given the assumption that the benefit cuts have the same present-
value impact on current generations under the two policies. The liabilities to existing
generations are the same in the aggregate, but the distribution among existing generations
differs, potentially significantly. The differences in generational burdens can also have
macroeconomic consequences if, for example, the consumption patterns and marginal
propensities to consume differ between the old and the young.
Similarly, government assets may have very different generational consequences,
some benefiting generations in the near term, others (for example, expenditures to reduce
greenhouse gas emissions) providing benefits in the distant future. If assets are included to
provide a more accurate picture of the generational consequences of fiscal policy, simply
including the value of assets does not provide a clear picture of the generational distribution of
benefits. Likewise, if we take account of the change in deferred tax assets associated with a
- 22 -
change in the tax structure, we will pick up the change in the deferred tax liabilities of current
generations, but not the distribution of that burden among these generations.
G. Deficit Targets and Budget Rules: Summary
Given what national debt and the budget deficit are presumed to tell us about the state
of fiscal policy, making adjustments to include government assets, implicit liabilities, and
deferred tax assets may be steps in the right direction, but these adjustments depend on the
specification of a benchmark fiscal system as well as additional assumptions and estimates,
given that legislation can change and that the assets and liabilities in question generally are
not traded and hence do not have established market values.
Even if these additional challenges are overcome, the resulting, more comprehensive
measure of national debt provides less information than we might wish about the full
generational consequences of fiscal policy. And, even with full information about the
distribution of benefits and the level of distortions, economic and demographic changes
present a challenge to determining what the optimal path of fiscal policy should be, because
the value of additional resource commitments changes over time.
Just as the appropriate measure of debt or the deficit is ambiguous because of these
various possible adjustments, the line between spending and revenues is defined, in the best of
circumstances, only by customary practice, and in many other cases is much less clear. This
presents a forceful challenge to the logic that spending targets are superior to deficit targets,
but leaves open the possibility of using both in the context of setting multiple targets for
policy.
4. Setting Long-Run Targets for Fiscal Policy
In light of the discussion in Sections 2 and 3, what targets should be used to assess the
government’s performance in relation to long-term fiscal policy objectives? It is useful to
- 23 -
distinguish the underlying objectives of policy, in terms of generational burdens, tax
smoothing, and sustainability, from the measures used to determine the performance of policy
with respect to these objectives. Presumably, the objectives should be set first, and the values
of the corresponding measures then determined. As stressed earlier, it is hard to see how one
can choose the desired path of a summary measure, such as the budget deficit, without first
determining the fundamental objectives of policy.
A. An Optimal Path for Debt and Deficits
Suppose, for the moment, that the structure of taxation and the structure of government
expenditures are fixed. This rules out policy changes that influence the composition of
expenditures or taxes, or that replace certain expenditure items with tax expenditures or vice
versa. This is the world implicitly assumed by many who assess fiscal policy using the deficit
as a summary measure and, while it represents a considerable abstraction from reality, it is a
good place to start. Let us also assume, initially, that there is no uncertainty about the
economy’s future path, once policy variables are chosen. What would optimal policy be in a
world like this, and how would optimal policy be characterized in terms of summary fiscal
targets?
Presumably, one would start with projections of cohort size and other demographic
characteristics relevant for assessing the valuation of government spending and tax payments
(e.g., birth rates, mortality profiles, morbidity, immigration rates, distribution of underlying
abilities etc.). Using this information, one would then consider all variations in annual
spending and annual taxes (the only two variables subject to choice in each year) that satisfied
the government’s intertemporal budget constraint, in each case assessing the generational
equity and economic efficiency of each combination to determine the resulting level of social
welfare. The dependence of social welfare on the intergenerational fiscal burden would
depend, of course, on one’s assumption about the strength of offsetting private
- 24 -
intergenerational transfers, and the impact of tax distortions would depend on both the tax
structure and the assumed behavioral responses to these distortions. The optimal fiscal policy
path would be the combination of annual spending and revenues yielding the highest level of
social welfare.
This optimal path could be fully characterized by a path of revenues and spending, or
alternatively by a path of deficits and spending (or by one of deficits and revenues). It could
not be fully characterized by the path of deficits alone, under the realistic assumption that
increases in taxes and reductions in spending do not have equivalent economic effects. Based
on a utilitarian principle that gives more weight to larger generations, it might be that
spending would be higher, and taxes lower, during periods when such generations were
affected by policy. To the extent that spending is more age-related than taxes, variations in
spending might more effectively accomplish this targeting of large generations, for example
causing a rise in old-age pension spending when large cohorts retired. And responses of
spending and deficits to changes in the population age structure would depend on the
permanence of demographic and economic changes. As discussed above, a permanent change
in the share of the population that is elderly would occasion a smaller increase in old-age
spending than a temporary change in the share, in order to provide for large future generations
of elderly as well.
Thus, we might see increases in spending and deficits as optimal when there are large
cohorts that benefit, with “large” being judged relative to the future rather than the past. On
the other hand, if spending related to large cohorts is anticipated only for the future, we would
expect lower current deficits, for the smoothing of taxes and burdens would require that the
cost of providing for these large cohorts be spread over all other cohorts, both before and
after.
- 25 -
What would this path of spending and deficits look like, given actual spending and
revenue patterns and demographic developments? Figure 1 provides an illustration for the
United States, based on calculations done by Auerbach, Furman and Gale (2007) using recent
long-term CBO projections.12 Auerbach et al. estimated, given CBO projections of revenues
and expenditures (excluding debt service) through 2085, and assuming constant revenue and
primary expenditures as a share of GDP thereafter, that the United States would require
increases in its primary surpluses equivalent to 6.01 percent of GDP annually in order to
satisfy the government’s intertemporal budget constraint. Without any change in policy the
deficit and debt would explode over time relative to GDP, as shown by the solid and dashed
red lines in Figure 1, with the debt-GDP ratio reaching over 500 percent by 2085.
The measured “fiscal gap” – in this case 6.01 percent of GDP – is, in itself, a useful
tool for summarizing the extent to which the current policy trajectory is unsustainable. Some
(e.g., Gokhale and Smetters 2003) have suggested expressing this fiscal gap as a present
value, to put it into the same units as the national debt. Doing so here results in a present-
value fiscal gap of $67.3 trillion, or about five times current GDP and about 14 times the net
outstanding national debt.
A major reason why the existing policy path is unsustainable is that, as the population
ages, old-age transfer programs for pensions and health care will absorb a greater share of
GDP. In addition, health care spending per beneficiary is projected to grow faster than GDP,
as it has in the past. The impact of these trends is evident in the increase of over 10 percent in
the expenditure-GDP ratio over the period.
Let us assume, for the sake of argument, that this pattern of expenditures is optimal,
i.e., that the increasing projected expenditures reflect increased valuation due to an older
population and a high value of marginal spending on health care. Then the optimal
12 An update of these projections may be found in CBO (2007b).
- 26 -
sustainable path would be one in which revenues were raised by the equivalent of 6.01 percent
per year. The solid and dashed black lines in Figure 1 show the resulting path of the deficit-
GDP and debt-GDP ratios, under the assumption that the optimal tax increase would occur
through a level increase in the revenue-GDP ratio, as might make sense based on both tax-
smoothing and generational incidence arguments. The revised policy calls for large budget
surpluses over the next several decades, in order to establish a fund to pay for subsequent
sustained high expenditures. As these surpluses accumulate, the national debt would quickly
vanish, with the debt-GDP ratio becoming more negative until just before the end of the
period shown in the figure.
Although this revised, sustainable path is based on apparently reasonable assumptions
regarding spending and revenues, it calls for a path for debt and deficits far from historical
values. Corresponding values for the primary surplus would reach as high as 8 percent of
GDP in the short term and would go negative starting only in 2050. Aside from whether the
path of expenditures and deficits really is optimal, this policy raises other economic and
political issues. First, how would a policy of retiring the national debt and accumulating
government assets be implemented, i.e., what assets would the government hold? This
question arose briefly in the United States early in 2001, when the forecast (prior to a series of
tax cuts, an acceleration of spending growth, and a recession) was for the national debt to
disappear. At that time, though, the forecast was for a rather short period without government
debt, so the question was less pressing than it would be if several decades of government asset
accumulation were in prospect. Second, how sustainable, politically, would a policy of debt
retirement followed by asset accumulation be? The presumption here is that the underlying
fiscal policy path is optimal, in which case the path of deficits and spending would be, as well.
But with policy traditionally using the annual deficit and debt-GDP ratio as a signal of
- 27 -
performance, substantial learning on the part of political actors and voters might be required
to accept large tax increases to pay for future benefits.
Finally, the policy path described, extending far into the future, corresponds to point
estimates of the relevant economic variables. But uncertainty with regard to these variables is
substantial and increases as the forecast horizon lengthens. For example, Figure 2 shows the
estimated five-year confidence intervals (with 95 percent confidence intervals in bold) for
deficit projections made by CBO in January, 2007, with projections five years out having a
95-percent confidence interval exceeding 10 percent of GDP. Clearly, point estimates of
policy paths convey only some of the relevant information when there is so much uncertainty
about the future.
With uncertainty present, the optimal fiscal policy path consists of a “tree” that
branches out over time rather than a single path. As new information arrives each year about
which branch the economy is on, the optimal path from that date forward must be computed.
But uncertainty also may affect the initial policy direction. For example, given the future
expenditure forecast in Figure 1, how much should the initial tax increase depend on the
degree of uncertainty attached to the expenditure forecast? One might have the intuition that
greater uncertainty, and in particular the possibility that the fiscal path can be sustained with
smaller or no tax increases than those based on point estimates, should lessen the need to
adopt deficit-reducing policies immediately. But such intuition is faulty because under usual
assumptions about household preferences, risk aversion pushes in the other direction, toward
societal precautionary saving, especially when there are constraints on policy changes and
even when there is a prospect that uncertainty will be resolved by waiting.13
Thus, even though one may expect to have to change policy frequently in the future as
a result of uncertainty, one cannot escape the need to make decisions now based on the
13 For further discussion and simulations, see Auerbach and Hassett (2007).
- 28 -
information available. But a planned path for debt and deficits should be accompanied by
confidence interval projections as well, to provide information on the magnitude of
uncertainty.
B. Government Assets, Implicit Liabilities and Deferred Taxes, Again
The discussion immediately preceding suggested that a planned trajectory for spending
and deficits could suffice as long-term fiscal policy targets. What, then, might be the role of
the various adjustments discussed above, involving the calculation of government assets,
implicit liabilities, and deferred tax assets? There are at least two arguments for including
these adjustments.
First, the derivation of the optimal deficit and spending trajectories presumed that the
generational patterns of spending and revenues were fixed. But variations in these patterns
are inevitable, in which case a given path of aggregate spending and revenue can correspond
to very different underlying patterns of economic distortions and generational burdens.
Incorporating the auxiliary adjustments for assets and liabilities, one can control to some
extent for these variations in generational patterns.
For example, suppose that government spending consists of two components, an age-
based public pension system plus general spending on public services, and that government
revenues consist of an income tax to pay for general spending plus a payroll tax on workers to
pay for the public pension system. In this example, an increase in spending on the public
pension system would shift the distribution of government spending toward the elderly, and
the pattern of government revenues toward workers. But computing the pension system’s
implicit liability, adding this liability to the explicit government debt, and changing revenues
and spending over time to be consistent with this change in the public debt measure, would
offset these changes in generational patterns by excluding payroll taxes and subsequent
benefit payments from the revenue and spending measures. The same rationale would apply
- 29 -
to an adjustment for deferred tax assets as a means for offsetting changes in the timing and
generational burden of tax collections and to an adjustment for government assets as an offset
to changes in the generational pattern of benefit from public spending.
A second argument for making these adjustments relates to the timing of
commitments. Consider, again, the public pension system, and contrast it to a situation in
which the government uses payroll taxes each year to finance expenditures that include
transfer payments to the elderly. Even though the generational consequences of this tax-
transfer system may be similar to that of the public pension system, the commitment of future
transfers to current workers is weaker in this case. To account for generational burdens, we
might wish to count implicit liabilities here, too. But to measure when commitments are
made, we might distinguish this case from the previous one. In a world of certainty and
commitment, in which the government chooses a sustainable path and then stays on it, there
would be no reason to treat the cases differently. But when there is both economic and
political uncertainty, it would seem to matter whether a plan to make transfers to retirees in
the future represents a firm commitment or just a plan. A stronger commitment may make the
fiscal trajectory less sustainable and may increase the expected transfer to the elderly, and
there should be some reflection of this in when liabilities are recognized.
Were we to include implicit liabilities in the national debt, and the changes in implicit
liabilities in the annual deficit, the adjustments could be quite large. In the United States, for
example, the unfunded liability to current participants in the public pension system was
estimated to be $16.5 trillion as of January 1, 2007 (OASDI Trustees, 2007, Table IV.B7).
The previous year, the unfunded liability was estimated to be $15.1 trillion (OASDI Trustees,
2006, Table IV.B7). The debt thus exceeds one-year’s GDP, which was $13.2 trillion in
2006, and the associated 2006 deficit of $1.4 trillion was over 10 percent of GDP. On the
other hand, such adjustments would make optimal deficit paths like the one derived in Figure
- 30 -
1 look much less strange to the untrained observer. The optimal policy would no longer call
for huge surpluses to pay for future spending if the timing of the spending were shifted to the
present.
Measures of implicit liabilities may fluctuate considerably from year to year simply
because of changes in forecasts of interest rates and demographic variables. For example,
Auerbach (2003) found that annual changes in economic and demographic assumptions could
change the estimated implicit deficit of the OASDI system by almost as much as the deficit
itself.14 There is nothing wrong with such fluctuations, but they might overstate the true
fluctuations in implicit liabilities to the extent that policy is flexible, e.g., if increases in
measured implicit liabilities make it more likely that future benefits will be reduced.
In summary, adjusting measured liabilities and spending for implicit liabilities,
government assets and deferred tax assets has some appeal as a method of making a given
year’s debt or deficit more meaningful, but there is no simple method for deciding precisely
which adjustments to make. This is particularly useful to keep in mind when the focus shifts
from fiscal targets to fiscal rules and restrictions, when simple rules may provide an
advantage.15
C. Generational Accounts and Related Measures
As described in Auerbach, Gohkale and Kotlikoff (1992) and elsewhere, generational
accounts distribute the burdens and benefits of annual taxes and spending to different
generations based on assumed patterns of incidence. The standard approach is to use taxes
and spending based on an assumed current-policy trajectory to calculate the present-value
accounts for generations currently alive, and to adjust the accounts for future generations
14 For example, the estimated deficit for 1998 would have been $581 billion using the 1997 economic and demographic assumptions to calculate the 1997 and 1998 implicit liabilities, but just $173 billion incorporating the changes in economic and demographic assumptions between the 1997 and 1998 forecasts.
15 One can imagine, for instance, perverse incentives for a government seeking to exclude implicit liabilities from the national debt by introducing gratuitous uncertainty about the likelihood of future benefit receipt.
- 31 -
away from current policy to the extent that current policy does not satisfy the government’s
intertemporal budget constraint. That is, in the terminology of the previous discussion, if
there is a fiscal gap, then the gap is assigned to future generations, and the difference between
the resulting burdens on future generations and those on current newborn generations, once
adjusted for trend growth, represents an alternative way of expressing the fiscal gap.16 By
dividing generational accounts by the present value of lifetime earnings, one obtains a path of
estimated lifetime average tax rates which, for any given tax structure, conveys information
about the tax distortions associated with fiscal policy.
Generational accounts provide measures that have a much closer connection to the
underlying objectives of government policy than do either adjusted or unadjusted measures of
deficits or spending. One potential argument against their use is the additional information
required for their calculation, as it is necessary not just to calculate annual flows of spending
and revenues, but to distribute these among generations. This argument may have some merit
in a context where estimates must be done rapidly, for example in a legislative setting where
new pieces of legislation are constantly being considered. But the argument carries less
weight when applied to calculations carried out less frequently, for example on an annual
basis as part of a fiscal policy evaluation. Indeed, generational accounts already have become
a tool for government fiscal policy evaluation in a number of countries.
A second potential argument against the use of generational accounts relates to the
previous discussion regarding whether to count implicit liabilities. Generational accounts
typically allocate all future taxes and spending along the assumed policy trajectory, except
where the spending is deemed difficult to assign to any particular generation. This procedure
automatically gives the same weight to all components of future taxes and spending,
16 Some critics of generational accounting have argued incorrectly that the methodology assumes that the fiscal gap will be borne entirely by future generations. The calculation simply expresses the fiscal gap in one particular way, as the extra burden that would be imposed on future generations if current policy held for all existing generations.
- 32 -
regardless of the strength of the commitments associated with these elements of fiscal policy.
A transfer payment made with certainty this year to a particular cohort is treated as equivalent
to a transfer payment with equal present value that is projected to be made to that cohort
twenty years hence, and two equal-size transfer payments projected to be made twenty years
hence are treated the same even if one has a stronger political or legal claim than the other. A
response to this criticism is to calculate generational accounts for a variety of plausible policy
scenarios, taking into account variations that might take place to provide a fuller picture of the
fiscal situation.
A third possible argument against using generational accounts is that they are less
familiar to policy makers than traditional deficit measures, or even measures of implicit
liabilities, so that they might be taken less seriously as indicators of the fiscal situation. This
has led some to suggest alternative ways of summarizing the generational accounts, for
example by reporting the present value of the liability being passed on to future generations
(Gokhale and Smetters 2003), but it is not clear why one would wish to reduce the amount of
information conveyed, particularly if generational accounts are provided in conjunction with
other fiscal policy measures such as those already discussed above.
D. Further Issues
All of the measures of fiscal policy discussed in this section are forward-looking and
involve uncertainty and the need to update projections on a regular basis. They also, in
principle, should be estimated using a model of the economy. That is, the behavioral
responses to policy changes should be taken into account, and the full impact of these
responses should be calculated using a general equilibrium model of the economy.
Ultimately, this is the only to way to determine whether a particular trajectory for spending
and deficits is socially optimal. That is, we can evaluate the generational burdens of a
particular policy and the economic distortions associated with a policy, but the only way to
- 33 -
weigh these effects precisely is in the context of an explicit model. This being said, the
specification and solution of such a model would require a massive number of simplifying
assumptions, many of which undoubtedly would be controversial and would lack strong
empirical foundation. This is not an argument against carrying out the general equilibrium
analysis, but it suggests that the basic “static” calculations should be presented as well and
that sensitivity analysis is important to shed light on the importance of different modeling
assumptions.17
One final issue is how any of the long-term measures discussed here should be
adjusted for short-term economic fluctuations. As discussed above, the current budget deficit
is automatically affected by economic fluctuations, and short-run stabilization policy might
dictate additional changes in spending, taxes and the deficit. Again, the optimal policy path,
taking short-term fluctuations into account, can in principle be identified using a
comprehensive general equilibrium model of the economy, but incorporating short-term
fluctuations into such a model involves yet more assumptions and controversy.18 Short of
such a model-based determination of optimal short-run policy fluctuations, one is left with the
standard approaches to deciding how much to allow budget targets to fluctuate in the short
run, but recognizing, consistent with a long tradition in the economics literature largely
ignored in the construction of budget rules, that different components of the budget have
different macroeconomic consequences.
5. Conclusions
Setting long-term targets for fiscal policy should start with a specification of
fundamental policy objectives. This would seem an obvious point but deserves emphasis
given that deficit targets often seem to appear through a process not unlike spontaneous
17 This was the approach taken by CBO (2006) when it evaluated the economic and budgetary effects of proposed changes in taxes and spending.
18 Indeed, the CBO (2006) exercise used different models to assess short-term and long-term effects of policy.
- 34 -
generation. The optimal policy plan should, in theory, consist of a specified trajectory, under
each possible course of the economy, of the different components of spending and taxes with
specified burdens on different generations. In practice, such a complete specification is
impossible, and the various summary measures discussed above can be very useful in
maintaining fiscal policy on a path that is sustainable, equitable and structured to promote
economic efficiency. There is no simple formula for how to compute the ideal trajectory, nor
is there a single measure of the economy’s fiscal situation that best addresses all issues. But a
collection of forward-looking measures, presented in conjunction with an assessment of their
dependence on particular assumptions, can provide far more information than short-term
deficit targets alone.
References
Alesina, Alberto, and Guido Tabellini, 1990, “A Positive Theory of Fiscal Deficits and Government Debt,” Review of Economic Studies 57:3, July, 403-414.
Altonji, Joseph G., Fumio Hayashi and Laurence J. Kotlikoff, 1992, “Is the Extended Family
Altruistically Linked? Direct Tests Using Micro Data,” American Economic Review 82:5, December, 1177-1198.
Altshuler, Rosanne, Nicholas Bull, John Diamond, Tim Dowd and Pamela Moomau, 2005,
“The Role of Dynamic Scoring in the Federal Budget Process: Closing the Gap between Theory and Practice,” American Economic Review 85:2, 432-436.
Auerbach, Alan J., 1983, “Corporate Taxation in the U.S.,” Brookings Papers on Economic
Activity, 14:2, 1451-1505. Auerbach, Alan J., 1996, “Tax Reform, Capital Allocation, Efficiency and Growth,” in H.
Aaron and W. Gale, eds., Economic Effects of Fundamental Tax Reform, Washington DC, Brookings, 29-81.
Auerbach, Alan J., 2003, “Fiscal Policy, Past and Present,” Brookings Papers on Economic
Activity, 34:1, 75-138. Auerbach, Alan J., 2005, “Dynamic Scoring: An Introduction to the Issues,” American
Economic Review 95:2, May, 421-425. Auerbach, Alan J., 2006a, “Budget Windows, Sunsets, and Fiscal Control,” Journal of Public
Economics 90:1/2, January, 87-100.
- 35 -
Auerbach, Alan J., 2006b, “The Choice between Income and Consumption Taxes: A Primer,” NBER Working Paper No. 12307, June.
Auerbach, Alan J., Jason Furman and William Gale, 2007, “Still Crazy after All These Years:
Understanding the Budget Outlook,” Tax Notes, May 21, 765-778. Auerbach, Alan J., William Gale and Peter Orszag, 2004, “The U.S. Fiscal Gap and
Retirement Saving,” OECD Economic Studies 39, 9-23. Auerbach, Alan J., Jagadeesh Gokhale and Laurence Kotlikoff, 1992, “Generational
Accounting: A New Approach to Understanding the Effects of Fiscal Policy on Saving,” Scandinavian Journal of Economics 94: 2, June, 303-318.
Auerbach, Alan J., and Kevin Hassett, 2007, “Optimal Long-Run Fiscal Policy: Constraints,
Preferences and the Resolution of Uncertainty,” Journal of Economic Dynamics and
Control 31:5, May 2007, 1451-1472. Auerbach, Alan J., and Laurence Kotlikoff, 1987, Dynamic Fiscal Policy, Cambridge,
Cambridge University Press, 1987. Auerbach, Alan J., and Ronald D. Lee, 2006, “Notional Defined Contribution Pension
Systems in a Stochastic Context: Design and Stability,” NBER Working Paper No. 12805, December.
Barro, Robert J., 1974, “Are Government Bonds Net Wealth?” Journal of Political
Economy.82:6, November, 1095-1117. Barro, Robert J., 1979, “Barro, Robert J., 1974, “On the Determination of the Public Debt,”
Journal of Political Economy 87:5, Part 1, October, 940-971. Board of Trustees, Federal Old-Age and Survivors Insurance and Federal Disability Insurance
Trust Funds, 2006, Annual Report. Board of Trustees, Federal Old-Age and Survivors Insurance and Federal Disability Insurance
Trust Funds, 2007, Annual Report. Bohn, Henning, and Robert Inman, 1996, “Balanced-Budget Rules and Public Deficits:
Evidence from the U.S. States, Carnegie-Rochester Conference Series on Public Policy 45, 13-76.
Congressional Budget Office, 1998, Emergency Spending Under the Budget Enforcement Act,
December. Congressional Budget Office, 2006, An Analysis of the President’s Budgetary Proposals for
Fiscal Year 2007, March. Congressional Budget Office, 2007a, The Uncertainty of Budget Projections: A Discussion of
Data and Methods, March. Congressional Budget Office, 2007b, The Long-Term Budget Outlook, December.
- 36 -
Eisner, Robert, and Paul J. Pieper, 1984, “A New View of the Federal Debt and Budget
Deficits,” American Economic Review 74:1, 11-29. Fair, Ray C., 1978, “The Effect of Economic Events on Votes for President,” Review of
Economics and Statistics 60:2, April, 159-173. Gokhale, Jagadeesh, and Kent Smetters, 2003, Fiscal and Generational Imbalances: New
Budget Measures for New Budget Priorities, Washington, D.C., The AEI Press. Persson, Torsten, and Lars E. O. Svensson, 1989, “Why a Stubborn Conservative would Run
a Deficit: Policy with Time- Inconsistent Preferences,” Quarterly Journal of
Economics 104:2, May, 325-345. Poterba, James M., 1994, “State Responses to Fiscal Crises: The Effects of Budgetary
Institutions and Politics,” Journal of Political Economy 102:4, August, 799-821. Rogoff, Kenneth, and Anne Sibert, 1988, “Elections and Macroeconomic Policy Cycles,”
Review of Economic Studies 55:1, January, 1-16. Sargent, Thomas J., and Neil Wallace, 1981, “Some Unpleasant Monetarist Arithmetic,”
Federal Reserve Bank of Minneapolis Quarterly Review 5:3, Fall. Surrey, Stanley S., 1974, Pathways to Tax Reform. Cambridge, Harvard University Press. Woodford, Michael, 1995, “Price Level Determinacy without Control of a Monetary
Aggregate,” Carnegie-Rochester Series on Public Policy 43, 1-46.
Ta
ble
1. T
he
Imp
act
of
Ch
an
ges
in
a P
ub
lic
Pen
sio
n S
yst
em
Imp
act
on
:
Exp
erim
ent
(1)
Cu
rren
t
Deb
t an
d
Def
icit
(2)
Sh
ort
-Ter
m
Deb
t an
d
Def
icit
(3)
Imp
lici
t
Lia
bil
itie
s
(4)
Old
er
Gen
erati
on
s
(5)
You
nger
Gen
erati
on
s
(6)
Fu
ture
Gen
erati
on
s
1.
Incr
ease
pay
roll
tax
es a
nd
asso
ciat
ed b
enef
its
by
sam
e pre
sent
val
ue,
st
arti
ng i
n f
ive
yea
rs
0
↓
0
0
0
0
2.
Red
uce
futu
re b
enef
its
foll
ow
ing s
ame
annual
pat
tern
as
tax
incr
ease
s in
ex
per
imen
t #1
0
↓
↓
↓
↓
↑
3.
Red
uce
futu
re b
enef
its
by
the
sam
e pre
sent
val
ue
as
in e
xper
imen
t #2, st
arti
ng
in 2
0 y
ears
0
0
↓
0
↓
↓
↑
Fig
ure
1. U
.S. E
xp
end
itu
res
an
d D
efic
its
-20
-100
10
20
30
40 2
007
201
22
01
720
22
20
27
20
32
203
72
04
220
47
20
52
20
57
206
22
06
720
72
20
77
20
82
Yea
r
Expenditures and Deficits, Percent of GDP
-30
0
-15
0
015
0
30
0
45
0
60
0
Debt, Percent of GDP
Ex
pen
dit
ure
s
Def
icit
, C
BO
Def
icit
, su
stai
nab
le p
ath
Deb
t, C
BO
Deb
t, s
ust
ain
able
pat
h
Fig
ure
2. D
efic
it C
on
fid
ence
In
terv
als
, J
an
ua
ry 2
00
7
-8-6-4-20246 19
89
19
91
19
93
19
95
19
97
19
99
20
01
20
03
20
05
20
07
20
09
20
11
Fis
cal
Yea
r
Percent of GDP
So
urc
e: C
BO
(2
00
7a)
Related Documents
